A Prototype Performance Analysis for V2VCommunications using USRP-based Software

Defined Radio Platform(Invited Paper)

Fei Peng, Shunqing Zhang, Shan Cao, and Shugong Xu,Shanghai Institute for Advanced Communication and Data Science

Key laboratory of Specialty Fiber Optics and Optical Access NetworksJoint International Research Laboratory of Specialty Fiber Optics and Advanced Communication

Shanghai University, Shanghai 200444, ChinaE-mail: {pfly shmily, shunqing, cshan, shugong}@shu.edu.cn

Abstract—Autonomous driving is usually recognized as apromising technology to replace human drivers in the nearfuture. To guarantee the safety performance in the daily lifescenario, multiple-car intelligence with high quality inter-vehiclecommunication capability is necessary in general. In this paper,to figure out the potential practical issues in the vehicle-to-vehicletransmission, we present a software defined radio platform forV2V communication using universal software radio peripheral(USRP). Based on the LTE framework, we modify the framestructure, the signal processing mechanisms and the resourceallocation schemes to emulate the updated LTE-V standard andgenerate the corresponding numerical results based on the realmeasured signals. As shown through some empirical studies, oneto four dB back-off is in general required to guarantee thereliability performance for V2V communication environments.

Index Terms—V2V communication, software defined radio,LTE-V, USRP.

I. INTRODUCTION

Autonomous driving [1], as a promising approach to replacehuman drivers, has attracted significant research attention fromindustry and academy recently, especially after the explosivedevelopment of artificial intelligence. With the assistance frommultiple sensors (camera, laser or millimeter wave radar),single-car intelligence provides acceptable performance in thesafety aspect. However, as reported in [2] and [3], the mostchallenging issue for autonomous driving is to provide guaran-teed safety performance under all scenarios, where multiple-car intelligence or intelligent connected vehicles becomescritical and the high quality inter-vehicle communication willbe necessary.

In order to support fast and reliable inter-vehicle communi-cation, traditional network oriented solution, such as long-termevolution (LTE), incurs notable transmission overhead betweenbase stations and vehicle terminals, where the direct transmis-sion among different vehicles will be a key enabler. Althoughthe traditional dedicated short range communications (DSRC)solutions provides vehicle-to-vehicle (V2V) communicationcapability, the quality-of-service cannot be guaranteed, which

triggers the new development of cellular-based vehicle-to-everything (V2X) solution such as LTE-vehicle (LTE-V) [4].In the current literature, various kinds of resource allocationschemes have been proposed to improve the reliability anddelay performance of LTE-V systems [5]. For instance, in [6],[7] and [8], the wireless link level performance improvementusing spectrum sharing, adaptive link assignment and poweradjustment are reported respectively. In addition, the path loss,shadow fading and delay spread of the V2V channels havebeen analyzed in [9]. Although the above results provide acomprehensive study in the numerical model based optimiza-tion framework, the following practical issues have NOT yetbeen investigated based on our present state of knowledge.

• Real-time Throughput and Reliability Evaluation In thetraditional model based evaluation method, the signalprocessing delay and reliability is usually “assumed” tobe a constant number, and the error rate performance is afixed value under given signal-to-noise (SNR) conditionin general. However, in the practical vehicular communi-cation scenario, the mobility induced channel variationsand the processing complexity definitely impact the real-time throughput and reliability performance, which iscritical for high reliable V2V transmission requirement.

• Hardware Imperfectness Another important factor to im-prove the V2V communication reliability is the pre-estimation of potential hardware impairments. Due tothe high mobility of vehicles, the Doppler spread in thewireless environment incurs the highly dynamic receivingsignals. As a result, the synchronization, the frequencyoffsets and the hardware mismatch issues become severe,which may impact the overall V2V transmission perfor-mance.

In this paper, we present a software defined radio (SDR)platform for V2V communication using universal softwareradio peripheral (USRP). Based on the LabVIEW LTE frame-work develop by National Instrument (NI) [10], we modifythe frame structure, the signal processing mechanisms and

arX

iv:1

809.

1041

3v2

[ee

ss.S

P] 2

Oct

201

8

the resource allocation schemes to emulate the updated LTE-V standard and generate the corresponding numerical resultsbased on the real measured signals. As shown through someempirical studies, we need a 1 to 4 dB transmit power back-off in order to guarantee the reliable block error rate (BLER)performance for V2V direct communication.

The rest of this paper is organized as follows. Section II pro-vides a preliminary introduction of V2V communication andthe associated standard progress. We demonstrate a softwaredefined radio platform which supports V2V communication inSection III and provides the corresponding emulation resultsin Section IV. Conclusions are given in Section V.

II. PRELIMINARY FOR LTE-V SYSTEMS

In this section, we provide a historical view of LTE-Vrelated standard. In order to support device-to-device (D2D)communication in the cellular framework, the third genera-tion partnership project (3GPP) proposes to introduce a newphysical link, called sidelink in release 12, which allows twouser equipments (UEs) to communicate directly without thesupport from base stations. However, to enable the direct V2Vcommunication based on the sidelink concept, there are stillmany practical issues in the vehicular network environment.

To address the above practical issues, the distributed re-source allocation mechanisms are first introduced in 3GPP re-lease 14, where the frame structure is modified to support sub-channelization of the previous sidelink implementation andthe energy sensing scheme is utilized before the direct V2Vtransmissions. In addition, the reference signal assignmentis modified according to support the potential high mobilityfrom V2V communication. On top of that, LTE-V use sensingbased semi-persistent scheduling in the higher layer to reducepotential collisions.

III. SDR PLATFORM FOR V2V TRANSMISSION

In this section, we provide an overview of current platformand the corresponding modifications for V2V transmission. Inaddition, we also describe some enhanced features for V2Vperformance evaluation and further improvement.

A. Platform Overview

From the hardware point of view, the SDR platform containsa programmable baseband processing unit and a reconfigurableradio frequency peripheral as shown in Fig. 1. To be morespecific, the LabVIEW LTE framework is realized through afield programmable gate array (FPGA) based programmablebaseband processing unit and a universal software radio pe-ripheral (USRP) based radio frequency part, which covers theoperating frequencies from 1.2 GHz to 6 GHz. In the softwareconfiguration, the initial version supported by LabVIEW LTEframework is 3GPP release 10, and the sidelink channel tosupport direct D2D/V2V transmission is required.

B. Adaptation to V2V Transmission

As summarized in Section II, to support V2V direct trans-mission on top of 3GPP release 10 framework, we need tomodify several processing parts as listed below.

Fig. 1. An overview of the software defined radio platform. In this platform,an FPGA-based programmable baseband processing unit and USRP-basedradio frequency part is configured in the prototype system, where the videotransmission applications are running through the air interfaces.

1) Frame Structure and Reference Signal: To facilitate di-rect V2V transmission, the sidelink frame structure is modifiedto jointly process the control and data information in somespecific symbols of each sub-frame according to 3GPP release14 specification [11], where the traditional approach to detectcontrol information before data processing is no longer used.On the contrary, the whole sub-frame will be buffered andthe corresponding control and data information is proceededthereafter as shown in Fig. 2. Through this approach, eachvehicle terminal can check the transmission status of eachsub-frame before transmitting and the distributed resourceallocation can be realized.

Fig. 2. An illustration example of 3GPP release 14 frame structure for V2VTransmission, where the first and last symbols are used for automatic gaincontrol and guard period respectively. In each sub-frame, four columns ofDMRSs are inserted for channel estimation purpose and the remaining partsare used for V2V transmission. According to the size of data, one or severalconsecutive sub-channels can be selected to transmit data.

Another important modification for V2V communicationis the reference signal assignment. In order to reduce theimpact of Doppler spread for high mobility vehicles, 3GPPrelease 14 [11] has specified to use up to four demodulationreference signals (DMRSs)1 per sub-frame. In addition, LTE-V systems also reserve the first symbol for automatic gain

1To be more specific, DMRSs are assigned in the 3rd, 6th, 9th and 12th

symbols of each sub-frame respectively.

control saturation and the last symbol for guard period. Notethat the resource allocation in V2V transmission is selectedon a sub-channel basis, where 48 subcarriers2 are grouped asa basic configuration of each sub-channel in the frequencydomain.

2) Information Processing and Resource Allocation: Inthe V2V transmission as explained before, the informationprocessing and resource allocation is utilized on the sub-channel basis rather than the traditional resource block basis.In this sense, the calculation of resource elements for controland data information per sub-frame is quite different fromthe conventional LTE transmission. Meanwhile, since differentnumber of consecutive sub-channels can be combined together,the control information extraction needs to be done per sub-channel basis and the data information can only be processedafter the number of consecutive sub-channels has been final-ized. As a result, the signal processing can be quite differentfrom the conventional LTE solution.

In the prototype system as shown in Fig. 3, we modifythe resource allocation part to support consecutive sub-channelassignment, where we allow random or pre-configured sub-channel assignment in each selection period3. In addition, weonly allow adjacent mode for resource assignment, e.g. thephysical sidelink control channel (PSCCH) and the physicalsidelink shared channel (PSSCH) are transmitted subsequently.

Fig. 3. Block diagram of baseband processing in the SDR platform. Themodification of frame structure and reference signal is mainly related tothe module of SL IQ Processing in both transmitting and receiving part.The modification of information processing and resource allocation is mainlyrelated to the module of PSCCH TX and PSCCH RX.

C. Advanced Features

To provide more insightful results and visualize the V2Vtransmission performance, we also introduce several advancedfeatures in what follows.

2For illustration purpose, we choose n to be 2, and the extension to otherstandard values of n is straight forward. We refer readers to [12] for moreinformation.

3In this paper, we choose the selection period to be 1ms or 10ms andthe extension to other standard compatible periods, such as 20ms, 50ms and100ms, will be straight forward.

1) Performance Evaluator: The first advanced feature thatincorporated in this prototype system is the performanceevaluator, which provides the possibility to track the systemperformance under different scenarios. Although to track theperformance of real-time prototype systems is quite challeng-ing in general, we design a fast caching scheme to temporallystore the decoding results of V2V transmission and calculatethe error rate performance in an offline manner. In this way, weare able to generate the reliability and throughput performanceaccordingly.

2) Connection with NS-3: Another important feature for theprototype system is the interaction capability with higher layersimulators. To make the V2V network evaluation feasible, weutilize the application programming interface on the LabVIEWplatform to receive the upper layer traffic conducted by the net-work simulator version 3 (NS-3) [13]. Through this approach,we are able to combine the LabVIEW physical layer with theupper layer communication process from NS-3, which allowsa more reliable system level solution design and performancevalidation.

IV. EMULATION RESULTS

In this section, we provide some empirical results basedon this SDR based V2V communication prototype system.To understand the imperfectness from wireless transmission,we build the indoor environment as shown in Fig. 1, wherethe V2V communication is modeled through two separatedantennas. Based on the LTE-V standard, we calculate theBLER and throughput of PSSCH as performance metrics.Other emulation parameters are summarized in Table I.

TABLE IEMULATION SETTINGS FOR SDR BASED V2V COMMUNICATION

PROTOTYPE SYSTEM

Parameter Value

Frequency 5.9 GHz

Modulation and Coding Scheme Level 0,5,10,15

Subchannel size 288 subcarriers

TX Power -20 to -6 dBmW

Inter-Antenna Distance 15 cm

Fading Environment Indoor

A. Reliability

In this experiment, we choose BLER versus transmit powerrelation to be the reliability measure. Different from thetraditional deterministic software simulation, we calculateBLER performance in each second (with fixed transmit power)and repeat this experiment by 4000 times in order to getthe statistics (average and standard deviation values) of thisrelation.

The average and standard deviation of BLER performanceversus transmit power relation under different modulation andcoding schemes (MCS) levels are shown in Fig. 4. Intuitively,a higher transmit power results in smaller BLER value on

average and a higher MCS level requires a higher transmitpower to maintain the similar reliability performance in gen-eral. However, from the standard deviation point of view, weare not able to obtain the similar monotonical relation. Ifwe further compare the ratios between the average value andstandard deviation of BLER under MCS level 10, the overallresults become 0.25 and 17.1 for the transmit power equalto -20 dBmW and -6 dBmW, respectively. In this sense, theaverage BLER may become unstable when the transmit powerincreases and in order to obtain a more reliable result, otherstatistics of BLER shall also be considered.

Fig. 4. Average BLER and standard deviation versus transmit power relationfor PSSCH transmission under different MCS levels. From this figure, wecan observe that: the average BLER decrease monotonically with respect tothe transmit power, while the standard deviation does not. By comparing theratios between the standard deviation and average BLER under MCS level10, we show that not only the average BLER value but also the statistics ofBLER shall also be considered when the transmit power increases.

Fig. 5. BELR versus transmit power relation for PSSCH transmission underdifferent MCS levels.

In Fig 5, we also compare the average BLER performancewith the BLER value under 99% confidence level. As wecan see from this figure, a 1 to 4 dB SNR boosting will benecessary to ensure a reliable BLER result.

B. ThroughputIn the second experiment, we plot the average throughput

versus MCS level relation to understand the achievable trans-mission rate of V2V communication. As shown in Fig. 6,

we are not able to obtain a monotonical relation betweenthe average throughput and MCS levels when the packetdrop happens in the high MCS levels. In other words, MCSadaptation is still necessary in the V2V communications. To bemore specific, the non-safety messages with lower reliabilityrequirement are more suitable for higher MCS levels, whilethe safety messages with higher reliability requirement needto trade off for lower MCS level.

In the practical V2V communications, MCS adaptationwould be more challenging, as it requires jointly consider thetarget BLER performance, the average throughput as well asthe reliability requirement, where additional research effortsare worthwhile.

Fig. 6. Average throughput of PSSCH versus different MCS level indices.The average throughput value of PSSCH channel increases monotonically inthe lower MCS level region, while it drops rapidly when the MCS level isabove 21. In the practical V2V communications, MCS adaptation would bemore challenging, as it requires jointly consider the target BLER performance,the average throughput as well as the reliability requirement, where additionalresearch efforts are worthwhile.

V. CONCLUSION

In this paper, we present an SDR based V2V communica-tion platform using the LabVIEW LTE framework. In orderto support direct V2V transmission, we modify the framestructure, the signal processing mechanisms and the resourceallocation schemes to emulate LTE-V systems. According tothe numerical results, we find that practical wireless envi-ronment and hardware imperfectness may result in a 1 to 4dB performance loss in terms of BLER if a higher reliabilityneeds to be guaranteed. In the future work, possible schemeswill be designed to improve the error performance, and morepractical scenarios will be considered for example makingthe experiment on moving vehicles. Meanwhile, we believethis platform will pave the way for future evaluation of V2Vnetworks and 5G V2X communication technologies.

ACKNOWLEDGEMENT

This work was supported by the National Natural ScienceFoundation of China (NSFC) Grants under No. 61701293, No.61871262, the National Science and Technology Major ProjectGrants under No. 2018ZX03001009, the Huawei InnovationResearch Program (HIRP), and research funds from ShanghaiInstitute for Advanced Communication and Data Science(SICS).

REFERENCES

[1] J. Baber, J. Kolodko, T. Noel, M. Parent, and L. Vlacic, “Cooperativeautonomous driving: intelligent vehicles sharing city roads,” IEEERobotics & Automation Magazine, vol. 12, no. 1, pp. 44–49, Mar. 2005.

[2] W. Li, X. Ma, J. Wu, K. S. Trivedi, X.-L. Huang, and Q. Liu, “Analyticalmodel and performance evaluation of long-term evolution for vehiclesafety services,” IEEE Transactions on Vehicular Technology, vol. 66,no. 3, pp. 1926–1939, Mar. 2017.

[3] L. Gao, Y. Li, J. Misener, and S. Patil, “C-V2X based basic safety relatedITS spectrum requirement analysis,” in IEEE Vehicular TechnologyConference (VTC-Fall), Sep. 2017.

[4] S. Chen, J. Hu, Y. Shi, and L. Zhao, “LTE-V: A TD-LTE-BasedV2X Solution for Future Vehicular Network,” IEEE Internet of ThingsJournal, vol. 3, no. 6, pp. 997–1005, Dec. 2016.

[5] J. Mei, K. Zheng, L. Zhao, Y. Teng, and X. Wang, “A Latency andReliability Guaranteed Resource Allocation Scheme for LTE V2V Com-munication Systems,” IEEE Transactions on Wireless Communications,vol. 17, no. 6, pp. 3850–3860, Jun. 2018.

[6] P. Wang, B. Di, H. Zhang, K. Bian, and L. Song, “Cellular V2XCommunications in Unlicensed Spectrum: Harmonious Coexistence withVANET in 5G systems,” IEEE Transactions on Wireless Communica-tions, vol. 17, no. 8, pp. 5212–5224, Aug. 2018.

[7] J. Lianghai, A. Weinand, B. Han, and H. D. Schotten, “Multi-RATssupport to improve V2X communication,” in IEEE Wireless Communi-cations and Networking Conference (WCNC), Apr. 2018.

[8] W. Sun, D. Yuan, E. G. Strm, and F. Brnnstrm, “Resource Sharing andPower Allocation for D2D-based safety-critical V2X communications,”in IEEE International Conference on Communication Workshop (ICCW),Jun. 2015.

[9] K. Guan, D. He, B. Ai, D. M. Matolak, Z. Zhong, and T. Kurner, “5-GHzObstructed Vehicle-to-Vehicle Channel Characterization for Internet ofIntelligent Vehicles,” to appear in IEEE Internet of Things Journal,2018.

[10] “Labview communications system design suite - national instruments,”http://www.ni.com/labview-communications/zhs/.

[11] 3rd Generation Partnership Project; Technical Specification GroupRadio Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA); Physical layer procedures (Release 15), 3GPP, Dec. 2017, ts36.213.

[12] 3rd Generation Partnership Project; Technical Specification GroupRadio Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA); Radio Resource Control (RRC); Protocol specification (Release15), 3GPP, Jan. 2018, ts 36.331.

[13] “Network simulator 3 (ns3),” https://www.nsnam.org/.